Recently Google Labs released it’s Audio Indexing initiative that indexing the spoken word within YouTube clips… using it’s speech recognition software. The tool is currently focused on Politician’s speeches, in particular the two most prominent politicians of the last few months, Obama and McCain. Here’s an example of a search for the word economy within this clip…
A couple of observations:
- The ability to search through indexed speech is going to add a whole new dimension to journalists and researchers attempts to finding relevant content. Can you imagine when this gets put out in Beta within Google’s Universal search? I guess to many it’ll be great to dig into that content but could it blurr the organic listings with irrelevant data? eg. indexed song lyrics that are not relevant to the actual search?
- Previously Google would index clips based on the meta-tags sitting behind the file, which means it was pretty much reliant on the honesty of the webmaster coding up the meta-tags. This move ensures that video files are indexed based on the actual spoken word… within each file. It will also encourage a whole load of marketing companies, bloggers and publishers to create and share more video content through YouTube and VLogs like Viddler and Seesmic.
Very cool technology… I’d love to see Google indexing the BBC or AlJazeera video archives…
No Comments on "Video and Speech Recognition…"